Now is the era of Voice User Interface (VUI) devices and they play a very important role as personal assistants. Here we compare the SUSI AI, Mycroft AI and Amazon Alexa based on the number of skills, their availability, easiness to add and edit skills and the provision of the user to modify the skill and add more to it if needed, etc.
- Starting with the number of skills, here Amazon Alexa supports way more number of skills as compared to both Mycroft AI and SUSI AI.
- Availability: Mycroft AI and SUSI AI are available everywhere and can set up anywhere regardless of the country whereas Alexa is available in U.S., U.K., Germany, India but they are aggressively expanding.
- Adding and editing skills: Mycroft and SUSI are open source and their skills can be added and edited and viewed by the open source community. Issues can be made to enhance the functionality of the skills whereas Alexa skills are not open source and certification and publishing of the skill is done by the Amazon team. Mycroft and SUSI skills can be customized by the user but this fails with Alexa as users have to create that same skill from scratch if they have to customize them.
- Platforms supported: Mycroft, SUSI and Alexa all support Linux. Mycroft lacks support for Windows and Mac but supports Raspberry Pi and Android, Alexa provides support for Windows and Mac and Raspberry Pi. SUSI also provides support for Android and iOS and can be integrated with speakers, vehicles, Pi, etc.
- Dedicated devices: As of now SUSI AI lacks such device. Mycroft has Mark 1 and Alexa has Echo. These devices are portable and are good candidates for home automation.
Due to different languages used, Mycroft AI skills can’t be directly used in SUSI AI. We need to convert Mycroft skills to SUSI skills if Mycroft skills are to be used for SUSI.
Some suggestions for making a dedicated device for SUSI:
- We can use a Raspberry Pi, USB headphones and a microphone to make a basic platform.
- We can install Jasper to enable the voice input on the Pi. Jasper is a open source application that enables us to make voice controlled applications.
- We can use SUSI server to interact with the device and the home appliances like lights. SUSI server can process the states of the the appliance (lights in this case) and return it as JSON objects to Raspberry Pi and then it may change the state as per user input.
Make a simple Hello World skill with SUSI:
- Visit https://github.com/fossasia/susi_skill_cms/blob/master/docs/Skill_Tutorial.md for a basic introduction to SUSI skills syntax and how does it work.
- Go to http://dream.susi.ai .
- Enter the skill name, say “hello”.
- You will be greeted by a welcome message – “roses are red…..”. Delete it and replace it with the following snippet.
|::name <Skill_name> #<— Enter skill name. for example hello
::author_url <author_url> #<— You can leave this empty as of now.
::description <description> #<— skill description
::developer_privacy_policy <link> #<— you can leave this as of now.
::image <image_url> #<— You can leave this as of now.
#Intent. Comments are written with a #
hi|hello|what’s up #<— This is what the user says
Hi|I am good|Hello #<— This is what the skill answers
6. Now go to http://susi.ai/chat for the testing.
7. In the SUSI chat dialog box (present at the bottom of the page) enter dream <test application name> where “test application name” is the name you enter when you first visit http://dream.susi.ai. In this case “dream hello”.
8. You can input “what’s up” in the dialog box and it will give you the desired output which you mentioned in the application.
SUSI has its own good points but it lacks in some department like the number and type of skills. Like Mycroft we can start making various skills and try to make a basic prototype of a dedicated SUSI personal assistant device.